Pith. sign in

REVIEW 3 major objections 5 minor 81 references

Rescuing double robustness: safe estimation under complete misspecification

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Doubly robust estimators can fail catastrophically when both nuisance models are wrong; clipping the correction term keeps the answer between two simpler estimates.

desk verdict A simple, well-motivated clipping fix for double robustness that is safe by construction; the bootstrap inference and the efficiency claims need work before the paper is publishable. read the letter →

arxiv 2509.22446 v2 pith:OJGBRRLJ submitted 2025-09-26 stat.ME math.STstat.MLstat.TH

Lorenzo Testa , Francesca Chiaromonte , Kathryn Roeder This is my paper · ORCID
classification stat.MEmath.STstat.MLstat.TH MSC 62D1062G2062F40
keywords doublyrobustestimationcompletemisspecificationdoublefragilityadaptivecorrectionclippingsafetyguaranteesemiparametricefficiencyparametricbootstrapmissingdata
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Doubly robust estimators are supposed to protect against misspecification, but the protection only works when at least one of the two nuisance models is correct. When both the outcome regression and the propensity score are wrong, the correction term can multiply their errors instead of cancelling them, producing estimates far worse than either simple component—a failure the paper calls double fragility. The paper's proposal, adaptive correction clipping (DR+ACC), redefines the doubly robust estimate by clipping the correction term so that the final estimate always lies between the outcome-regression and inverse-probability-weighted estimates. This preserves consistency and semiparametric efficiency when one model is correct, while guaranteeing under complete misspecification an error no larger than the better of the two component estimators. The paper also provides a parametric bootstrap for confidence intervals and demonstrates the method on the Kang–Schafer benchmark and on Alzheimer's proteomics data.

What carries the argument

The central object is the adaptive correction clipping operator $\mathrm{clip}(\hat C;\hat\theta_{OR},\hat\theta_{IPW})$, which replaces the raw doubly robust correction $\hat C=n^{-1}\sum_i R_i\hat\mu(X_i)/\hat\pi(X_i)$ by its truncation to the interval whose endpoints are the OR and IPW estimates. The interval property—that $\hat\theta_{DR+ACC}$ is always a convex combination of $\hat\theta_{OR}$ and $\hat\theta_{IPW}$—carries the argument: it converts the product-of-errors remainder of standard DR into a convex combination and max bound of component errors, and it preserves consistency whenever one component is correct because clipping is continuous and the unclipped correction converges to the boundary in that scenario. The machinery also includes an asymptotic expansion that passes the clip through the limit, yielding the non-Gaussian limiting distribution used by the parametric bootstrap.

What would settle it

Repeat the Kang–Schafer simulation with both nuisance models correctly specified but estimated on the estimation sample rather than pre-trained, and measure the empirical coverage of the parametric-bootstrap intervals at $n=200$ and $n=1000$. If coverage falls materially below 95%, the same-sample covariance estimate is not consistent and the inference claim does not extend to the paper's simulation protocol.

Watch

Extended reading notes

Core claim

The paper claims that the celebrated double robustness of estimators like $\hat\theta_{DR}=\hat\theta_{OR}+\hat\theta_{IPW}-\hat C$ is best understood as asymptotic hard thresholding: when at least one nuisance model is correct the correction term $\hat C$ cancels the misspecified component, but when both nuisance models are wrong the correction compounds their errors. The proposed estimator $\hat\theta_{DR+ACC}=\hat\theta_{OR}+\hat\theta_{IPW}-\mathrm{clip}(\hat C)$, with $\mathrm{clip}(\hat C)$ truncated to $[\min(\hat\theta_{OR},\hat\theta_{IPW}),\max(\hat\theta_{OR},\hat\theta_{IPW})]$, keeps the doubly robust consistency property (Theorem 3.2) and is safe: for every sample its error is at most the maximum of the errors of the OR and IPW estimators (Theorem 3.4). Because the clipping map is nonlinear, the limiting distribution is the nonlinear transform $Z_{OR}+Z_{IPW}-\mathrm{clip}(Z_{C})$, not Gaussian; the paper proves that a parametric bootstrap based on the joint influence-function covariance gives asymptotically valid intervals when both nuisances are well specified (Theorem 3.7). A complete replication of the Kang–Schafer design shows the unclipped DR estimator's RMSE exploding from about 2.6 to 21.9 at $n=200$ and from 1.1 to 77.6 at $n=1000$ under complete misspecification, while DR+ACC stays near the better component, and the Alzheimer's application finds 40 additional significant peptides beyond the standard DR analysis.

Load-bearing premise

The theory assumes the nuisance models are pre-trained or estimated on a separate sample; when they are estimated on the same sample used to build the confidence intervals, the covariance estimate behind the bootstrap has no established validity.

Editorial extensions

If this is right

  • If at least one nuisance model is correctly specified, DR+ACC stays consistent and has the same semiparametric efficiency as the standard DR estimator, so adopting it costs nothing in the ideal case.
  • If both nuisance models are wrong, a user who trusts the better of their OR and IPW estimates is guaranteed that DR+ACC cannot be worse than that better component on any sample—whereas the unclipped DR estimate can be many times worse (RMSE 77.6 vs 1.7 at $n=1000$ in the benchmark).
  • When both nuisance models are correct, the parametric bootstrap produces intervals with coverage close to nominal, so inference is not lost by giving up asymptotic normality.
  • Because the method only clips a prespecified correction, arbitrary black-box nuisance models—neural networks or large language models—can be plugged in without changing the statistical guarantees.
  • In the Alzheimer's proteomics application, the safety property changes conclusions: DR+ACC finds 95 significant peptides at the 5% level versus 55 with standard DR, and 34 of the extra genes have independent literature support.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same interval-truncation trick could be applied to any doubly robust correction term for other estimands—average treatment effects, distribution functions, policy effects—turning it into a generic safe-debiasing wrapper.
  • The paper's own simulations estimate nuisances on the estimation sample even though the theory assumes pre-trained models; formally proving the parametric bootstrap under cross-fitting would close this gap.
  • A natural testable extension is to compare DR+ACC against propensity trimming and self-normalized estimators under targeted overlap violations, since the safety bound holds pointwise but the efficiency ranking under partial misspecification beyond the benchmark scenario is unexplored.
  • The 34 AD-related genes that DR+ACC newly flags should be treated as a hypothesis for replication in independent proteomic cohorts, not as confirmed discoveries.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper studies doubly robust (DR) estimation of a missing-data mean when both nuisance models—the outcome regression and the propensity score—may be misspecified. It first documents the phenomenon of 'double fragility': under complete misspecification the correction term of a DR estimator can amplify, rather than remove, bias. It then proposes DR+ACC, which clips the correction term so that the final estimate always lies between the outcome-regression (OR) and inverse-probability-weighting (IPW) estimates. The paper proves consistency when at least one nuisance model is correct (Theorem 3.2), a deterministic safety bound (Theorem 3.4), and an asymptotic distribution result for the clipped estimator under correct specification (Theorem 3.7), together with a parametric bootstrap procedure (Algorithm 1). The method is evaluated in simulations replicating Kang and Schafer (2007) and in an Alzheimer's disease proteomics application.

Significance. If the inference claims were fully supported, DR+ACC would be a practically useful and conceptually clean solution to a real fragility problem: it preserves the double-robustness property when the theory applies and guarantees, by construction, that the estimate stays within the interval spanned by the two simpler estimators. The safety theorem is correct, and the paper is commendably transparent about the non-normal limit and about the bootstrap rather than the normal approximation. The simulations and the real-data application give rich empirical support for the point-estimation benefits. However, the central inference claim—valid confidence intervals via the parametric bootstrap—is not established for the data-dependent nuisance setting used in the simulations and application, and the abstract's assertion of 'no reduction in semiparametric efficiency' is not proved and appears at odds with the non-normal limit in Theorem 3.7. These issues are load-bearing because the paper's practical conclusions (the proteomics findings, the coverage tables) rely on them.

major comments (3)
  1. [Section 3.2, Algorithm 1, Theorem 3.7] The inference procedure is not justified when the nuisance functions are estimated from the data. The asymptotic expansion in Eq. (20) and the covariance matrix estimator in Algorithm 1 are appropriate only when μ̂ and π̂ are fixed and pre-trained, as assumed in the theoretical setup of Section 2.1. When μ̂ and π̂ are estimated on the same sample, θ̂_OR and θ̂_IPW are not asymptotically linear with influence functions μ*(X)−θ* and RY/π*(X)−θ*; the nuisance-estimation error contributes first-order terms to the joint distribution, and the same applies to the correction term. The paper states in Section 2.1 that the results 'can be readily extended' to sample splitting, but no theorem, rate condition, or corrected variance formula is supplied. Sections 4 and 5 estimate the nuisances on the estimation sample itself, with no sample splitting, so the coverage results in Table 2 and the significance findings in Section 5 rest on an inference procedure whose validity is not established by the paper's theory.
  2. [Abstract and Section 3.2, Theorem 3.7] The abstract claims that the proposal comes with 'no reduction in semiparametric efficiency' compared with DR estimators, but no efficiency theorem is proved and the claim appears false as stated. Theorem 3.7 gives the limit W = Z_OR + Z_IPW − clip(Z_correction), which differs from the efficient DR limit Z_OR + Z_IPW − Z_correction whenever Z_correction falls outside the interval [min(Z_OR,Z_IPW), max(Z_OR,Z_IPW)]; this event has positive probability in general. The paper only remarks (Remark 3.8 and Supplementary Figure C.1) that the difference is negligible in simulations. The authors should either prove that the asymptotic variance (or a suitable efficiency criterion) is unchanged, with explicit computation of the probability of clipping, or remove the efficiency claim from the abstract and conclusions.
  3. [Theorem 3.4 and Remark 3.5] The safety statement is essentially a restatement of the clipping construction rather than a derived property. Equation (17) defines λ̂ as the weight that makes θ̂_DR+ACC = λ̂ θ̂_OR + (1−λ̂) θ̂_IPW, so the first inequality in (18) is an algebraic identity given the interval property proved in the three cases. This is not a flaw in the construction, but the abstract's wording that the error is 'bounded by a convex combination of the individual nuisance model errors' overstates the result: Theorem 3.4 bounds the estimator error by a convex combination of the errors of the OR and IPW estimators, not directly of the nuisance functions. In addition, Remark 3.5's bias bound (19) requires λ̂ to be estimated on an independent sample, but the paper's procedure does not do this; the practical relevance of (19) is therefore unclear.
minor comments (5)
  1. [Eq. (14) and Lemma A.1] The notation 'clip' is used with one argument in Eq. (14) and Algorithm 1, but with three arguments in Lemma A.1 and Lemma A.3. This makes the definition of the clipping operator in the bootstrap simulation, where the bounds are the simulated min/max of Z_OR and Z_IPW, ambiguous.
  2. [Theorem 3.4, Eq. (17)] The definition of λ̂ has a zero denominator when θ̂_OR = θ̂_IPW. The degenerate case should be handled explicitly (e.g., by defining λ̂ = 1/2 or by a limiting argument), even if it has probability zero under continuity.
  3. [Section 6] There is a typo in the Conclusions: 'posesses' should be 'possesses'.
  4. [Section 5] The statements about '55 peptides' and '95 significant peptides' are made without multiple testing correction, and the claim that 34 genes were 'independently implicated' in Alzheimer's disease is based on a literature list rather than a statistical validation. The text should be more careful to present these as exploratory and hypothesis-generating.
  5. [Section 3.2, Eq. (20)] The notation in (20) uses Z_correction both for a limiting random variable and for the scaled error of the unclipped correction term; this is acceptable but could be clarified by defining the joint convergence of the three-vector explicitly.

Circularity Check

1 steps flagged · score 2.0 of 10

Safety theorem restates the clipping construction; consistency and bootstrap results are independent.

  1. self definitional [Section 1.1 (estimator construction) and Section 3.1, Theorem 3.4]
    "Our method works by constraining, or “clipping”, the correction term to ensure the final estimate is always anchored within the range defined by the simpler OR and IPW estimates. ... Then, we have |θ̂DR+ACC−θ⋆| ≤ λ̂|θ̂OR−θ⋆| + (1−λ̂)|θ̂IPW−θ⋆| ≤ max{|θ̂OR−θ⋆|, |θ̂IPW−θ⋆|}."

    The DR+ACC estimator is defined as θ̂OR + θ̂IPW − clip(C), with the clip explicitly forcing the correction term into the interval [min{θ̂OR, θ̂IPW}, max{θ̂OR, θ̂IPW}]. Consequently, by construction θ̂DR+ACC is a convex combination of θ̂OR and θ̂IPW, and Theorem 3.4 merely restates this interval membership as an error bound via the triangle inequality. The safety guarantee is hardwired into the estimator's definition rather than derived from independent assumptions or data. This is a transparent tautology rather than a hidden circular argument: the paper states the design goal before proving the theorem. The consistency and asymptotic-distribution results are not definitional and carry the paper's independent content.

full rationale

The only reduction-by-construction located in the derivation chain is the safety theorem, which is a direct restatement of the clipping construction: the estimator is explicitly designed to remain between the OR and IPW estimates, so the max-error bound follows without additional statistical input. This is a mild, transparent design property, not a fitted parameter renamed as a prediction and not a load-bearing self-citation. The paper's substantive results—consistency under partial misspecification (Theorem 3.2), the non-Gaussian asymptotic limit (Theorem 3.7), and the parametric bootstrap confidence procedure (Algorithm 1)—are derived from component expansions and asymptotic arguments and do not assume the safety conclusion. The bootstrap covariance issue under estimated nuisance functions is a correctness/robustness gap, not a circularity: the paper asserts without proof that sample splitting extends the results, and the simulations estimate nuisances on the estimation sample, but that does not make the conclusion an input of the derivation. Accordingly, the overall circularity score is low.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The DR+ACC estimator has no tuning parameters. Analysis choices in the application (10% missingness cutoff, MissForest imputation, covariate list) are data-processing decisions, not fitted parameters of the method. The paper introduces no new physical or probabilistic entities. 'Double fragility' is a label for an existing empirical phenomenon, not a new postulated object.

assumptions (5)
  • domain assumption Missing at random: R ⊥⊥ Y | X (Assumption 2.1a).
    Needed for identifiability of θ* and for the influence function expansion in Lemma 2.2.
  • domain assumption Weak overlap: π(x) in (0,1) for all x almost surely (Assumption 2.1b).
    Ensures inverse probability weights are well defined and bounded, used throughout the proofs.
  • standard math Existence of regular asymptotically linear estimators and influence functions for the target parameter (Eq. 5).
    Basis for the semiparametric framework and for the asymptotic expansion in Theorem 3.7.
  • domain assumption At least one nuisance model is exactly well-specified for the consistency result (Theorem 3.2).
    This is the standard double robustness condition; the clipping must be asymptotically inactive for it to preserve consistency.
  • domain assumption Both nuisance models are well-specified and the scaled estimators have a joint normal limit with consistently estimable covariance for the bootstrap (Theorem 3.7, Algorithm 1).
    The inference section explicitly assumes well-specified nuisances; the consistency of Σ̂ is not proved when nuisances are estimated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rescuing double robustness: safe estimation under complete misspecification." pith.science (2026). https://pith.science/paper/OJGBRRLJ

@misc{pith2026250922446,
  author       = {Pith},
  title        = {Pith review of: Rescuing double robustness: safe estimation under complete misspecification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OJGBRRLJ}},
  note         = {Machine review of arXiv:2509.22446}
}
read the original abstract

Double robustness is a major selling point of semiparametric and missing data methodology. Its virtues lie in protection against partial nuisance misspecification and asymptotic semiparametric efficiency under correct nuisance specification. However, in many applications, complete nuisance misspecification should be regarded as the norm (or at the very least the expected default), and thus doubly robust estimators may behave fragilely. In fact, it has been amply verified empirically that these estimators can perform poorly when all nuisance functions are misspecified. Here, we first characterize this phenomenon of double fragility, and then propose a solution based on adaptive correction clipping (DR+ACC). We argue that our DR+ACC proposal is safe, in that it inherits the favorable properties of doubly robust estimators under correct nuisance specification, but its error is guaranteed to be bounded by a convex combination of the individual nuisance model errors, which prevents the instability caused by the compounding product of errors of doubly robust estimators. We also show that our proposal comes with no reduction in semiparametric efficiency compared to doubly robust estimators, and thus valid inference based on asymptotic normality can be conducted when nuisances are well-specified. We showcase the efficacy of our DR+ACC estimator both through extensive simulations and by applying it to the analysis of Alzheimer's disease proteomics data.

Figures

Figures reproduced from arXiv: 2509.22446 by the authors.

Figure 1
Figure 1. Sampling distributions for the Outcome Regression (OR), Inverse Probability Weighting (IPW), Doubly Robust (DR), and our proposed DR+ACC estimators from 1000 simulations with a sample size of n = 1000. The true parameter value is 210. The four panels show the estimators’ performance under all combinations of correct and incorrect nuisance model specifications. The top row and bottom-left panel demonstrate the asympt… view at source ↗
Figure 2
Figure 2. A conceptual illustration of the behavior of the doubly robust estimator under different model specification scenarios. Scenario A depicts the case of partial misspecification, where the OR estimator is consistent for θ ⋆ but the IPW estimator is biased. The DR estimator correctly aligns with the consistent OR estimator, demonstrating its asymptotic hard thresholding property. Scenario B depicts the case of complete… view at source ↗
Figure 3
Figure 3. Sampling distributions for the Outcome Regression (OR), Inverse Probability Weighting (IPW), Doubly Robust (DR), and our proposed DR+ACC estimators from 1000 simulations with a sample size of n = 200. The true parameter value is 210. The four panels show the estimators’ performance under all combinations of correct and incorrect nuisance model specifications. The top row and bottom-left panel demonstrate the asympto… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: In the left panel, we show the distributions of the estimated average treatment effects (ATEs) of Alzheimer’s disease across 270 different peptides. All estimators produce distributions centered near zero, suggesting that for most peptides, the estimated effect of AD i…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

81 extracted references · 51 canonical work pages

  1. [1]

    Leukotriene a4 hydrolase inhibition improves age-related cognitive decline via modulation of synaptic function

    Julia M Adams, Sanket V Rege, Angela T Liu, Ninh V Vu, Sharda Raina, Douglas Y Kirsher, Amy L Nguyen, Reema Harish, Balazs Szoke, Dino P Leone, et al. Leukotriene a4 hydrolase inhibition improves age-related cognitive decline via modulation of synaptic function. Science Advances, 9 0 (46): 0 eadf8764, 2023

    2023

  2. [2]

    Pcs-uq: Uncertainty quantification via the predictability-computability-stability framework

    Abhineet Agarwal, Michael Xiao, Rebecca Barter, Omer Ronen, Boyu Fan, and Bin Yu. Pcs-uq: Uncertainty quantification via the predictability-computability-stability framework. arXiv preprint arXiv:2505.08784, 2025

    arXiv 2025

  3. [3]

    Prediction-powered inference

    Anastasios N Angelopoulos, Stephen Bates, Clara Fannjiang, Michael I Jordan, and Tijana Zrnic. Prediction-powered inference. Science, 382 0 (6671): 0 669--674, 2023 a

    2023

  4. [4]

    Ppi++: Efficient prediction-powered inference

    Anastasios N Angelopoulos, John C Duchi, and Tijana Zrnic. Ppi++: Efficient prediction-powered inference. arXiv preprint arXiv:2311.01453, 2023 b

    arXiv 2023

  5. [5]

    Semi-supervised linear regression

    David Azriel, Lawrence D Brown, Michael Sklar, Richard Berk, Andreas Buja, and Linda Zhao. Semi-supervised linear regression. Journal of the American Statistical Association, 117 0 (540): 0 2238--2251, 2022

    2022

  6. [6]

    Doubly robust estimation in missing data and causal inference models

    Heejung Bang and James M Robins. Doubly robust estimation in missing data and causal inference models. Biometrics, 61 0 (4): 0 962--973, 2005

    2005

  7. [7]

    An essay on the logical foundations of survey sampling, part i

    D Basu. An essay on the logical foundations of survey sampling, part i. foundations of statistical inferences, vp godambe and da sprott, 1971

    work page 1971

  8. [8]

    Efficient and adaptive estimation for semiparametric models, volume 4

    Peter J Bickel, Chris AJ Klaassen, Ya’acov Ritov, and Jon A Wellner. Efficient and adaptive estimation for semiparametric models, volume 4. Springer, 1993

    work page 1993

Show all 81 references
  1. [9]

    The emerging role of peptidyl-prolyl isomerase chaperones in tau oligomerization, amyloid processing, and alzheimer's disease

    Laura J Blair, Jeremy D Baker, Jonathan J Sabbagh, and Chad A Dickey. The emerging role of peptidyl-prolyl isomerase chaperones in tau oligomerization, amyloid processing, and alzheimer's disease. Journal of neurochemistry, 133 0 (1): 0 1--13, 2015

    2015

  2. [10]

    C-learner: Constrained learning for causal inference

    Tiffany Tianhui Cai, Yuri Fonseca, Kaiwen Hou, and Hongseok Namkoong. C-learner: Constrained learning for causal inference. arXiv preprint arXiv:2405.09493, 2024

    2024 arXiv

  3. [11]

    Challenges of the inconsistency regime: Novel debiasing methods for missing data models

    Michael Celentano and Martin J Wainwright. Challenges of the inconsistency regime: Novel debiasing methods for missing data models. arXiv preprint arXiv:2309.01362, 2023

    2023 arXiv

  4. [12]

    Robust semi-parametric inference in semi-supervised settings

    Abhishek Chakrabortty. Robust semi-parametric inference in semi-supervised settings. PhD thesis, Harvard University, 2016

    2016

  5. [13]

    Efficient and adaptive linear regression in semi-supervised settings

    Abhishek Chakrabortty and Tianxi Cai. Efficient and adaptive linear regression in semi-supervised settings. The Annals of Statistics, 46 0 (4): 0 1541--1572, 2018

    2018

  6. [14]

    Bioinformatics methods for mass spectrometry-based proteomics data analysis

    Chen Chen, Jie Hou, John J Tanner, and Jianlin Cheng. Bioinformatics methods for mass spectrometry-based proteomics data analysis. International journal of molecular sciences, 21 0 (8): 0 2873, 2020

    2020

  7. [15]

    Reduced rna adenosine-to-inosine editing in hippocampus vasculature associated with alzheimer’s disease

    Philip S Crooke III, John T Tossberg, Rachel M Heinrich, Krislyn P Porter, and Thomas M Aune. Reduced rna adenosine-to-inosine editing in hippocampus vasculature associated with alzheimer’s disease. Brain Communications, 4 0 (5): 0 fcac238, 2022

    2022

  8. [16]

    Optimal and safe estimation for high-dimensional semi-supervised learning

    Siyi Deng, Yang Ning, Jiwei Zhao, and Heping Zhang. Optimal and safe estimation for high-dimensional semi-supervised learning. Journal of the American Statistical Association, 119 0 (548): 0 2748--2759, 2024

    2024

  9. [17]

    Neuropathy due to impaired axonal transport of non-fragmented mitochondria in myh14 mutation carriers

    Josef Finsterer. Neuropathy due to impaired axonal transport of non-fragmented mitochondria in myh14 mutation carriers. EBioMedicine, 49: 0 24, 2019

    2019

  10. [18]

    Finite population regression adjustment and non-asymptotic guarantees for treatment effect estimation

    Mehrdad Ghadiri, David Arbour, Tung Mai, Cameron Musco, and Anup B Rao. Finite population regression adjustment and non-asymptotic guarantees for treatment effect estimation. Advances in Neural Information Processing Systems, 36: 0 74180--74212, 2023

    2023

  11. [19]

    The s-value: evaluating stability with respect to distributional shifts

    Suyash Gupta and Dominik Rothenh \"a usler. The s-value: evaluating stability with respect to distributional shifts. Advances in Neural Information Processing Systems, 36: 0 72058--72070, 2023

    2023

  12. [20]

    The influence curve and its role in robust estimation

    Frank R Hampel. The influence curve and its role in robust estimation. Journal of the american statistical association, 69 0 (346): 0 383--393, 1974

    1974

  13. [21]

    Robust Statistics: The Approach Based on Influence Functions

    Frank R Hampel, Elvezio M Ronchetti, Peter J Rousseeuw, and Werner A Stahel. Robust Statistics: The Approach Based on Influence Functions. John Wiley & Sons, 2011

    2011

  14. [22]

    Distinguishing alzheimer’s disease patients and biochemical phenotype analysis using a novel serum profiling platform: potential involvement of the vwf/adamts13 axis

    Jay S Hanas, James RS Hocker, Christian A Vannarath, Megan R Lerner, Scott G Blair, Stan A Lightfoot, Rushie J Hanas, James R Couch, and Linda A Hershey. Distinguishing alzheimer’s disease patients and biochemical phenotype analysis using a novel serum profiling platform: pote...

    2021

  15. [23]

    Reweighted ls estimators converge at the same rate as the initial estimator

    Xuming He and Stephen Portnoy. Reweighted ls estimators converge at the same rate as the initial estimator. The Annals of Statistics, pages 2161--2167, 1992

    1992

  16. [24]

    Estimating causal effects from epidemiological data

    Miguel A Hern \'a n and James M Robins. Estimating causal effects from epidemiological data. Journal of Epidemiology & Community Health, 60 0 (7): 0 578--586, 2006

    2006

  17. [25]

    Alzheimer’s disease associated genes ankyrin and tau cause shortened lifespan and memory loss in drosophila

    James P Higham, Bilal R Malik, Edgar Buhl, Jennifer M Dawson, Anna S Ogier, Katie Lunnon, and James JL Hodge. Alzheimer’s disease associated genes ankyrin and tau cause shortened lifespan and memory loss in drosophila. Frontiers in cellular neuroscience, 13: 0 260, 2019

    2019

  18. [26]

    A generalization of sampling without replacement from a finite universe

    Daniel G Horvitz and Donovan J Thompson. A generalization of sampling without replacement from a finite universe. Journal of the American statistical Association, 47 0 (260): 0 663--685, 1952

    1952

  19. [27]

    Integrated analysis of differential gene expression profiles in hippocampi to identify candidate genes involved in alzheimer's disease

    Wanhua Hu, Xiaodong Lin, and Kelong Chen. Integrated analysis of differential gene expression profiles in hippocampi to identify candidate genes involved in alzheimer's disease. Molecular medicine reports, 12 0 (5): 0 6679--6687, 2015

    2015

  20. [28]

    Robust statistical procedures

    Peter J Huber. Robust statistical procedures. SIAM, 1996

    1996

  21. [29]

    Tbcd may be a causal gene in progressive neurodegenerative encephalopathy with atypical infantile spinal muscular atrophy

    Toshio Ikeda, Akihiko Nakahara, Rie Nagano, Maiko Utoyama, Megumi Obara, Hiroshi Moritake, Tamayo Uechi, Jun Mitsui, Hiroyuki Ishiura, Jun Yoshimura, et al. Tbcd may be a causal gene in progressive neurodegenerative encephalopathy with atypical infantile spinal muscular atroph...

    2017

  22. [30]

    Demystifying double robustness: A comparison of alternative strategies for estimating a population mean from incomplete data

    Joseph DY Kang and Joseph L Schafer. Demystifying double robustness: A comparison of alternative strategies for estimating a population mean from incomplete data. Statistical Science, 22 0 (4): 0 523--539, 2007

    2007

  23. [31]

    Semiparametric doubly robust targeted double machine learning: a review

    Edward H Kennedy. Semiparametric doubly robust targeted double machine learning: a review. Handbook of statistical methods for precision medicine, pages 207--236, 2024

    2024

  24. [32]

    Sharp instruments for classifying compliers and generalizing causal effects

    Edward H Kennedy, Sivaraman Balakrishnan, and Max G’Sell. Sharp instruments for classifying compliers and generalizing causal effects. The Annals of Statistics, 48 0 (4): 0 2008--2030, 2020

    2008

  25. [33]

    Semi-supervised u-statistics

    Ilmun Kim, Larry Wasserman, Sivaraman Balakrishnan, and Matey Neykov. Semi-supervised u-statistics. arXiv preprint arXiv:2402.18921, 2024

    2024 arXiv

  26. [34]

    Increased levels of stress-inducible phosphoprotein-1 accelerates amyloid- deposition in a mouse model of alzheimer’s disease

    Rachel E Lackie, Jose Marques-Lopes, Valeriy G Ostapchenko, Sarah Good, Wing-Yiu Choy, Patricija van Oosten-Hawle, Stephen H Pasternak, Vania F Prado, and Marco AM Prado. Increased levels of stress-inducible phosphoprotein-1 accelerates amyloid- deposition in a mouse model of ...

    2020

  27. [35]

    Rab gtpases, membrane trafficking and diseases

    Guangpu Li. Rab gtpases, membrane trafficking and diseases. Current drug targets, 12 0 (8): 0 1188--1193, 2011

    2011

  28. [36]

    Identification of novel risk genes for alzheimer’s disease by integrating genetics from hippocampus

    Jie Li, Lingfang Li, Shanshan Cai, Kun Song, and Shenghui Hu. Identification of novel risk genes for alzheimer’s disease by integrating genetics from hippocampus. Scientific Reports, 14 0 (1): 0 27484, 2024

    2024

  29. [37]

    Differentially expressed genes in alzheimer’s disease highlighting the roles of microglia genes including olr1 and astrocyte gene cdk2ap1

    Qingqin S Li and Louis De Muynck. Differentially expressed genes in alzheimer’s disease highlighting the roles of microglia genes including olr1 and astrocyte gene cdk2ap1. Brain, behavior, & immunity-health, 13: 0 100227, 2021

    2021

  30. [38]

    Binding of pro-inflammatory proteins s100a8 or s100a9 to amyloid- peptide suppresses its fibrillation

    Ekaterina A Litus, Marina P Shevelyova, Alisa A Vologzhannikova, Evgenia I Deryusheva, Andrey V Machulin, Ekaterina L Nemashkalova, Maria E Permyakova, Andrey S Sokolov, Valeria D Alikova, Vladimir N Uversky, et al. Binding of pro-inflammatory proteins s100a8 or s100a9 to amyl...

    2025

  31. [39]

    Neuronal expression of ilei/fam3c and its reduction in alzheimer’s disease

    Lei Liu, Naoki Watanabe, Hiroyasu Akatsu, and Masaki Nishimura. Neuronal expression of ilei/fam3c and its reduction in alzheimer’s disease. Neuroscience, 330: 0 236--246, 2016

    2016

  32. [40]

    Alzheimer's disease and aging association: identification and validation of related genes

    T Liu, K Hou, J Li, T Han, S Liu, and Jianshe Wei. Alzheimer's disease and aging association: identification and validation of related genes. The Journal of Prevention of Alzheimer's Disease, 11 0 (1): 0 196--213, 2024

    2024

  33. [41]

    Opposing effects of histone h2a

    Jian Qi Luo, Luca A Hategan, Samantha Creighton, Shan Hua, Timothy AB McLean, Tarkan Ahmad Dahi, Zhenhong Jin, Fardad Pirri, Stephen M Winston, Isaiah L Reeves, et al. Opposing effects of histone h2a. z on memory, transcription and pathology in male and female alzheimer’s dise...

    2025

  34. [42]

    Doubly robust estimators with weak overlap

    Yukun Ma, Pedro HC Sant'Anna, Yuya Sasaki, and Takuya Ura. Doubly robust estimators with weak overlap. arXiv preprint arXiv:2304.08974, 2023

    2023 arXiv

  35. [43]

    Lithium alters expression of rnas in a type-specific manner in differentiated human neuroblastoma neuronal cultures, including specific genes involved in alzheimer’s disease

    Bryan Maloney, Yokesh Balaraman, Yunlong Liu, Nipun Chopra, Howard J Edenberg, John Kelsoe, John I Nurnberger, and Debomoy K Lahiri. Lithium alters expression of rnas in a type-specific manner in differentiated human neuroblastoma neuronal cultures, including specific genes in...

    2019

  36. [44]

    A peptide-centric quantitative proteomics dataset for the phenotypic assessment of alzheimer’s disease

    Gennifer E Merrihew, Jea Park, Deanna Plubell, Brian C Searle, C Dirk Keene, Eric B Larson, Randall Bateman, Richard J Perrin, Jasmeer P Chhatwal, Martin R Farlow, et al. A peptide-centric quantitative proteomics dataset for the phenotypic assessment of alzheimer’s disease. Sc...

    2023

  37. [45]

    Augmented doubly robust post-imputation inference for proteomic data

    Haeun Moon, Jin-Hong Du, Jing Lei, and Kathryn Roeder. Augmented doubly robust post-imputation inference for proteomic data. bioRxiv, pages 2024--03, 2025

    2024

  38. [46]

    Off-policy estimation of linear functionals: Non-asymptotic theory for semi-parametric efficiency

    Wenlong Mou, Martin J Wainwright, and Peter L Bartlett. Off-policy estimation of linear functionals: Non-asymptotic theory for semi-parametric efficiency. arXiv preprint arXiv:2209.13075, 2022

    2022 arXiv

  39. [47]

    The heme degradation pathway is a promising serum biomarker source for the early detection of alzheimer's disease

    Claudius Mueller, Weidong Zhou, Amy VanMeter, Michael Heiby, Shino Magaki, Mark M Ross, Virginia Espina, Matthew Schrag, Cindy Dickson, Lance A Liotta, et al. The heme degradation pathway is a promising serum biomarker source for the early detection of alzheimer's disease. Jou...

    2010

  40. [48]

    Brain and csf alzheimer’s biomarkers are associated with serpine1 gene expression

    Cynthia Picard, Henrik Zetterberg, Kaj Blennow, Sylvia Villeneuve, Judes Poirier, and Prevent-Ad Research Group. Brain and csf alzheimer’s biomarkers are associated with serpine1 gene expression. Genes, 16 0 (7): 0 818, 2025

    2025

  41. [49]

    Interactions of cellular energetic gene clusters in the alzheimer’s mouse brain

    Raghavan Pillai Raju, Lun Cai, Alpna Tyagi, and Subbiah Pugazhenthi. Interactions of cellular energetic gene clusters in the alzheimer’s mouse brain. Molecular neurobiology, 61 0 (1): 0 476--486, 2024

    2024

  42. [50]

    Pcs workflow for veridical data science in the age of ai

    Zachary T Rewolinski and Bin Yu. Pcs workflow for veridical data science in the age of ai. arXiv preprint arXiv:2508.00835, 2025

    2025

  43. [51]

    Estimation of regression coefficients when some regressors are not always observed

    James M Robins, Andrea Rotnitzky, and Lue Ping Zhao. Estimation of regression coefficients when some regressors are not always observed. Journal of the American statistical Association, 89 0 (427): 0 846--866, 1994

    1994

  44. [52]

    A cell-autonomous role of dnajc5 in microglial-mediated proteostasis in alzheimer’s disease

    Matthew J Rosene, Niko-Petteri Nykanen, Logan Brase, Oscar Harari, and Bruno A Benitez. A cell-autonomous role of dnajc5 in microglial-mediated proteostasis in alzheimer’s disease. Alzheimer's & Dementia, 19: 0 e071075, 2023

    2023

  45. [53]

    Adjusting for nonignorable drop-out using semiparametric nonresponse models

    Daniel O Scharfstein, Andrea Rotnitzky, and James M Robins. Adjusting for nonignorable drop-out using semiparametric nonresponse models. Journal of the American Statistical Association, 94 0 (448): 0 1096--1120, 1999

    1999

  46. [54]

    Astrocyte transcriptomic changes along the spatiotemporal progression of alzheimer’s disease

    Alberto Serrano-Pozo, Huan Li, Zhaozhi Li, Clara Mu \ n oz-Castro, Methasit Jaisa-Aad, Molly A Healey, Lindsay A Welikovitch, Rojashree Jayakumar, Annie G Bryant, Ayush Noori, et al. Astrocyte transcriptomic changes along the spatiotemporal progression of alzheimer’s disease. ...

    2024

  47. [55]

    Neural epidermal growth factor-like like protein 2 is expressed in human oligodendroglial cell types

    Mohammed R Shaker, Amna Kahtan, Renuka Prasad, Ju-Hyun Lee, Giovanni Pietrogrande, Hannah C Leeson, Woong Sun, Ernst J Wolvetang, and Andrii Slonchak. Neural epidermal growth factor-like like protein 2 is expressed in human oligodendroglial cell types. Frontiers in Cell and De...

    2022

  48. [56]

    The er protein canx (calnexin)-mediated autophagy protects against alzheimer disease

    Hongtao Shen, Yuying Xie, Yan Wang, Yusheng Xie, Yongxiang Wang, Zhenyan Su, Laixi Zhao, Shi Yao, Xiaoling Cao, Jinglan Liang, et al. The er protein canx (calnexin)-mediated autophagy protects against alzheimer disease. Autophagy, 21 0 (5): 0 1096--1115, 2025

    2025

  49. [57]

    The moringin/ -cd pretreatment induces neuroprotection in an in vitro model of alzheimer’s disease: A transcriptomic study

    Serena Silvestro, Luigi Chiricosta, Agnese Gugliandolo, Renato Iori, Patrick Rollin, Daniele Perenzoni, Fulvio Mattivi, Placido Bramanti, and Emanuela Mazzon. The moringin/ -cd pretreatment induces neuroprotection in an in vitro model of alzheimer’s disease: A transcriptomic s...

    2021

  50. [58]

    Missforest—non-parametric missing value imputation for mixed-type data

    Daniel J Stekhoven and Peter B \"u hlmann. Missforest—non-parametric missing value imputation for mixed-type data. Bioinformatics, 28 0 (1): 0 112--118, 2012

    2012

  51. [59]

    Propensity score weighting and trimming strategies for reducing variance and bias of treatment effect estimates: a simulation study

    Til St \"u rmer, Michael Webster-Clark, Jennifer L Lund, Richard Wyss, Alan R Ellis, Mark Lunt, Kenneth J Rothman, and Robert J Glynn. Propensity score weighting and trimming strategies for reducing variance and bias of treatment effect estimates: a simulation study. American ...

    2021

  52. [60]

    Impact of c-reactive protein on cognition and alzheimer disease biomarkers in homozygous apoe e4 carriers

    Qiushan Tao, Ting Fang Alvin Ang, Samia C Akhter-Khan, Indira Swetha Itchapurapu, Ronald Killiany, Xiaoling Zhang, Andrew E Budson, Katherine W Turk, Lee Goldstein, Jesse Mez, et al. Impact of c-reactive protein on cognition and alzheimer disease biomarkers in homozygous apoe ...

    2021

  53. [61]

    Semiparametric semi-supervised learning for general targets under distribution shift and decaying overlap

    Lorenzo Testa, Qi Xu, Jing Lei, and Kathryn Roeder. Semiparametric semi-supervised learning for general targets under distribution shift and decaying overlap. arXiv preprint arXiv:2505.06452, 2025

    2025 arXiv

  54. [62]

    Semiparametric theory and missing data, volume 4

    Anastasios A Tsiatis. Semiparametric theory and missing data, volume 4. Springer, 2006

    2006

  55. [63]

    Targeted maximum likelihood learning

    Mark J Van Der Laan and Daniel Rubin. Targeted maximum likelihood learning. The international journal of biostatistics, 2 0 (1), 2006

    2006

  56. [64]

    Asymptotic statistics, volume 3

    Aad W Van der Vaart. Asymptotic statistics, volume 3. Cambridge university press, 2000

    2000

  57. [65]

    Causal inference: A statistical learning approach, 2024

    Stefan Wager. Causal inference: A statistical learning approach, 2024

    2024

  58. [66]

    Non-asymptotic bounds of aipw estimators for means with missingness at random

    Fei Wang and Yuhao Deng. Non-asymptotic bounds of aipw estimators for means with missingness at random. Mathematics, 11 0 (4): 0 818, 2023

    2023

  59. [67]

    Time-course global proteome analyses reveal an inverse correlation between a burden and immunoglobulin m levels in the appnl-f mouse model of alzheimer disease

    Hansen Wang, Declan Williams, Jennifer Griffin, Takashi Saito, Takaomi C Saido, Paul E Fraser, Ekaterina Rogaeva, and Gerold Schmitt-Ulms. Time-course global proteome analyses reveal an inverse correlation between a burden and immunoglobulin m levels in the appnl-f mouse model...

    2017

  60. [68]

    Pac style guarantees for doubly robust generalized front-door estimator

    Yuhao Wang, Arnab Bhattacharyya, Jin Tian, and NV Vinodchandran. Pac style guarantees for doubly robust generalized front-door estimator. In 9th Causal Inference Workshop at UAI 2024, 2024

    2024

  61. [69]

    Transcriptomic analysis of human brains with alzheimer’s disease reveals the altered expression of synaptic genes linked to cognitive deficits

    Jamal B Williams, Qing Cao, and Zhen Yan. Transcriptomic analysis of human brains with alzheimer’s disease reveals the altered expression of synaptic genes linked to cognitive deficits. Brain communications, 3 0 (3): 0 fcab123, 2021

    2021

  62. [70]

    Benchmarking estimators for natural experiments: A novel dataset and a doubly robust algorithm

    R Teal Witter and Christopher Musco. Benchmarking estimators for natural experiments: A novel dataset and a doubly robust algorithm. Advances in Neural Information Processing Systems, 37: 0 82594--82626, 2024

    2024

  63. [71]

    A unified framework for semiparametrically efficient semi-supervised learning

    Zichun Xu, Daniela Witten, and Ali Shojaie. A unified framework for semiparametrically efficient semi-supervised learning. arXiv preprint arXiv:2502.17741, 2025

    2025 arXiv

  64. [72]

    Albumin antagonizes alzheimer’s disease-related tau pathology and enhances cognitive performance by inhibiting aberrant tau aggregation

    Ming-Xuan Yang, Zhuo-Ran Wang, Yan-Li Zhang, Zhi-Na Zhang, Yan-Li Li, Rui Wang, Qiang Su, and Jun-Hong Guo. Albumin antagonizes alzheimer’s disease-related tau pathology and enhances cognitive performance by inhibiting aberrant tau aggregation. Experimental Neurology, 386: 0 1...

    2025

  65. [73]

    Genetic and expression analysis of copi genes and alzheimer’s disease susceptibility

    Yu Yang, Xu Wang, Weina Ju, Li Sun, and Haining Zhang. Genetic and expression analysis of copi genes and alzheimer’s disease susceptibility. Frontiers in Genetics, 10: 0 866, 2019

    2019

  66. [74]

    Nrbp1-containing crl2/crl4a regulates amyloid production by targeting bri2 and bri3 for degradation

    Takashi Yasukawa, Aya Tsutsui, Chieri Tomomori-Sato, Shigeo Sato, Anita Saraf, Michael P Washburn, Laurence Florens, Tohru Terada, Kentaro Shimizu, Ronald C Conaway, et al. Nrbp1-containing crl2/crl4a regulates amyloid production by targeting bri2 and bri3 for degradation. Cel...

    2020

  67. [75]

    Investigating pathogenetic mechanisms of alzheimer’s disease by systems biology approaches for drug discovery

    Shan-Ju Yeh, Ming-Hsun Chung, and Bor-Sen Chen. Investigating pathogenetic mechanisms of alzheimer’s disease by systems biology approaches for drug discovery. International Journal of Molecular Sciences, 22 0 (20): 0 11280, 2021

    2021

  68. [76]

    Stability

    Bin Yu. Stability. Bernoulli, 19 0 (4): 0 1484--1500, 2013

    2013

  69. [77]

    Veridical data science: The practice of responsible data analysis and decision making

    Bin Yu and Rebecca L Barter. Veridical data science: The practice of responsible data analysis and decision making. MIT Press, 2024

    2024

  70. [78]

    Veridical data science

    Bin Yu and Karl Kumbier. Veridical data science. Proceedings of the National Academy of Sciences of the United States of America, 117 0 (8): 0 3920--3929, 2020

    2020

  71. [79]

    Semi-supervised inference: General theory and estimation of means

    Anru Zhang, Lawrence D Brown, and T Tony Cai. Semi-supervised inference: General theory and estimation of means. The Annals of Statistics, 47 0 (5): 0 2538--2566, 2019

    2019

  72. [80]

    High-dimensional semi-supervised learning: in search of optimal inference of the mean

    Yuqian Zhang and Jelena Bradic. High-dimensional semi-supervised learning: in search of optimal inference of the mean. Biometrika, 109 0 (2): 0 387--403, 2022

    2022

  73. [81]

    Cross-prediction-powered inference

    Tijana Zrnic and Emmanuel J Cand \`e s. Cross-prediction-powered inference. Proceedings of the National Academy of Sciences, 121 0 (15): 0 e2322083121, 2024

    2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.